A Hybrid Method to Improve Forecasting Accuracy

概要

論文の詳細を見る
Formerly, following method was proposed by us. Focusing that the equation of exponential smoothing method was equivalent to (1,1) order ARMA model equation, the new method of estimation of smoothing constant in exponential smoothing method was proposed. In this paper utilizing the above stated method, the revised forecasting method is proposed. In making forecast such as sales data, seasonal trend is removed by dividing the original time series by the Monthly Ratio which is calculated using moving average data Utilizing the above stated method, forecasting is executed with this data in which seasonal trend is removed. The Monthly Ratio is multiplied after that, and forecasting data is obtained. This is a revised forecasting method. Variance of forecasting error of this newly proposed method is assumed to be less than those of previously proposed method. These method are examined by the data of airline passengers. These results are compared with those of ARIMA model. Some favorable results are obtained.
大阪府立大学の論文